# Qwen 3.7 Max

> Alibaba's newest closed-weight flagship. 1M context, top reasoning + multilingual.

Human version: https://kymaapi.com/models/qwen-3.7-max
Live JSON: `GET https://kymaapi.com/v1/models` (no auth required)

## Facts

- **Model ID**: `qwen-3.7-max` — pass this as `model` in the request body
- **Creator**: Alibaba
- **Released**: 2026-05-21
- **Context window**: 1M tokens
- **Max output**: 33K tokens per response — a hard ceiling, not a default
- **Price**: $2.56 in / $7.676 out per 1M, cached input 10%
- **Capabilities**: tools, reasoning, caching

## Call it

```bash
curl https://kymaapi.com/v1/chat/completions \
  -H "Authorization: Bearer $KYMA_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model": "qwen-3.7-max", "messages": [{"role": "user", "content": "Hello"}]}'
```

## Positioning

Alibaba's newest closed-weight flagship, carrying a 1M-token context window with top-tier reasoning and multilingual strength. Reach for it when the job is too large or too hard for everyday-tier models.

## About

Qwen 3.7 Max is Alibaba's newest closed-weight flagship — a premium model built for general work, reasoning, multilingual tasks, and long context. It is the model to pick when output quality matters more than cost per token.

On Kyma it runs through the same OpenAI-compatible endpoint as every other model, with automatic failover rerouting requests if a serving path degrades. Prompt caching is supported, so workloads that resend a long system prompt or document prefix pay 10% of the input rate on the repeated part — which matters at premium pricing. Every response reports its exact cost in usage.cost.

The 1M-token context window pairs with function calling, structured outputs, and extended reasoning, with up to 32K tokens of output per request. It is text-in, text-out only — no image inputs.

## Use cases

- **Hard reasoning problems** — Extended reasoning works through math, planning, and analysis tasks where step-by-step thinking decides the outcome.
- **Whole-corpus analysis** — The 1M context takes entire codebases, document sets, or transcript archives in one request instead of a retrieval pipeline.
- **Multilingual production work** — Translation, localization, and cross-language analysis where quality has to hold up beyond English.
- **High-stakes coding** — Code review, architecture decisions, and complex refactors where a premium model earns its rate.
- **Structured agent pipelines** — Function calling plus structured outputs keep multi-step agents reliable on long-horizon tasks.

## Not ideal for

Image inputs (it's text-only — pick a vision model), and high-volume or latency-sensitive workloads, where its premium pricing and medium speed make a cheaper, faster tier the better fit.

## Pick something else when

- You want a cheaper strong default → `qwen-3.6-plus`
- You need the strongest tool-heavy agent behavior → `kimi-k2.6`
- You want top reasoning at lower cost → `deepseek-v4-pro`

## See also

- All models: https://kymaapi.com/models.md
- Pricing: https://kymaapi.com/pricing.md
- Other models by Alibaba: https://kymaapi.com/models?q=Alibaba